Study on the current velocity prediction by Artificial Neural Network at the entrance of Hualien Port of Taiwan
Chih-Tsung Hsiao, Ching-Her Hwang · 2010 Sixth International Conference on Natural Computation · 2010
This study uses an Artificial Neural Network (ANN) model, and selects a set of time-series data, including wave height, wave period, and ocean current velocity, as observed by buoys at the marine meteorology observation station, mounted at a depth of -34 m in the sea outside of the entrance to Hualien Port, Taiwan, for a period of 9 days, from August 5 to 13, 2007. The data are used as a base for comparison and modification of the simulation analysis for port current velocity predictions. According to comparisons between the research results and the common time series AR (2) model analysis results, the root-mean-square errors of the predicted values of current velocity per second and the measured values of the two analytical models are 0.397 cm and 0.341 cm, respectively, indicating that the ANN model is feasible for predictions of ocean current velocity in ports.